At smbd.trans2.c:329, strcasecmp() is called,
but this should be StrCaseCmp().
strcasecmp() does not care of multibyte characters,
for example it recognize a Japanese character 0x8141
and 0x8161 is same one, but they should be different.

This affects that the 0x8161's long name becomes 
0x8141 (0x8161 is correct). To replace strcasecmp() 
to StrCaseCmp() solves the problem. Here is a patch.

--- smbd/trans2.c.org	Sun Jun 22 20:09:02 2003
+++ smbd/trans2.c	Thu Jun 26 02:27:59 2003
@@ -326,7 +326,7 @@
 		return False;
 	if (case_sig)	
 		return strcmp(str,mask)==0;
-	return strcasecmp(str,mask) == 0;
+	return StrCaseCmp(str,mask) == 0;
 }
 
 /****************************************************************************

This patch is found by MORIYAMA Masayuki <msyk@mtg.biglobe.ne.jp>